The images in this article were generated with artificial intelligence. How we publish
Cybersecurity researchers have described a patch-free vulnerability that allows an attacker to cause the filtration of a user's NTLMv2 hash by inducing the opening of manipulated links in the browser. Although mechanics recalls the incident solved in April 2026 with the URI handle of the Snipping Tool (ms-screenshot), the new technique takes advantage of the handle search: with type parameters crumb = location: to force the system to connect against a resource controlled by the attacker and thus capture the Net-NTLMv2 that the team tries to send to authenticate.
The root of the problem is the same type of failure: a protocol handler that accepts parameters provided by the user without validating them and that, when processing them, causes a outgoing connection to UNC routes. That connection to a malicious SMB server triggers the NTLM protocol and exposes the hash, which an attacker can use for relay or to try to authenticate within a compromised network. Previous cases had already shown how parameters such as filePath or crumb could be exploited for the same purpose; companies like Varonis have documented uses of crumb In 2024, and suppliers like Huntress have published analysis of recent derivations.

Microsoft's decision not to publish a patch for this finding on the argument that it did not reach its threshold of gravity for servicing leaves organizations in the face of a specific operational risk: in the absence of an official arrangement, vectors are still exploitable by actors who can attract users to malicious links embedded in mail or web pages. This makes the protection on the periphery of the network and the policies of control of protocol and port in workstations particularly relevant.
In terms of impact, the exposure of Net-NTLMv2 is dangerous because hash can serve as a basis for relay attacks and side movements, and because many corporate environments continue to allow or tolerate NTLM for compatibility with legacy software. A single has high privileges whose partial credential is filtered can facilitate climbing and persistence in a network, with consequences that include data exfiltration, malware deployment or internal service commitment.
As practical and viable immediate countermeasures, it is advisable to block outgoing SMB traffic (TCP / 445 and TCP / 139) in equipment that does not need to communicate with internal servers through these ports from public networks or unsafe subnetworks. In addition, forcing SMB signature and enabling extended authentication protection policies reduces the possibility of a captured hash being successfully reused against it. Where feasible, organizations should plan to phase out NTLM in favour of Kerberos and other safer authentication.
From the detection and response layer, you need to monitor unusual attempts at NTLM authentication to external destinations, review SMB connection records and deploy EDR rules that alert processes that launch URI handlers with external parameters. Staff awareness remains critical: social engineering techniques that induce the click remain the preferred vector, so training and filtering of sandboxing mail helps reduce the likelihood of exploitation.

At the strategic level, this incident recalls two lessons: first, that local protocol managers can become exit doors if they accept unvalidated entries; second, that the decisions of suppliers on what parking may leave gaps that require countermeasures at the level of configuration and architecture by security officials. Implementing network segmentation, egress restrictions and host firewall rules are measures that will mitigate exposure while waiting for a formal patch.
Those who want to deepen can read analysis and technical publications from the teams that have investigated variants of this attack on specialized blogs and official documentation about NTLM and SMB. Useful resources for contextualizing and planning mitigation include the input of manufacturers and community on NTLM and SMB management, such as Microsoft technical documentation on NTLM and the publications of security companies that have documented holdings with parameters in URI managers: https: / / learn.microsoft.com / en-us / windows-server / security / ntlm / and https: / / www.varonis.com / blog, in addition to reports and research blogs of specialized sites such as https: / / www.huntress.com / blog.
Finally, my recommendation for IT and safety managers is to prioritize controls that do not depend on a supplier patch: to block unnecessary outgoing SMB, strengthen SMB signature, plan NTLM deactivation and improve authentication monitoring. These measures reduce the opportunity window of the attackers and raise the operating cost even if the vulnerability remains uncorrected by the manufacturer.
Related
More news on the same subject.

GitLab critical alert: emergency patch fixes CVE-2026-19478 allowing to modify or eliminate public projects without credentials
GitLab published an emergency patch on August 17, 2026 to correct critical vulnerability in its self-hosted software (Community and Enterprise Edition) which, under certain cond...

When the MCP server keeps your credentials: the silent attack vector of the IA in production
The incorporation of IA agents into business processes has opened a practical way for production systems and data to be accessible from models: it is called Model Context Protoc...

Critical alert: CVE-2026-58231 in SAP Commerce Cloud could allow remote code execution; patch and urgent mitigation
A critical vulnerability that affects SAP Commerce Cloud, registered as CVE-2026-58231 and with maximum score 10.0 on the CVSS scale, it is being exploited attempts shortly afte...

The massive purchase of expired domains drives fraud, malware and streaming pirate: the business behind the dropcatch
An intelligence report on DNS published by Infoblox and disseminated by specialized media confirms that criminals are buying large-scale expired domains - the so-called dropcatc...

HoneyMyte updates CoolClient with a signed kernel driver to hide processes and protect the C2 channel
Kaspersky has published an analysis that attributes to the actor known as HoneyMyte (also Mustang Panda) an updated version of the CoolClient backdoor that incorporates a signed...

GeoServer on zero-day vulnerability alert in jsonArrayContains with real risk of remote execution
The GeoServer open source project has a zero-day vulnerability that is being actively explored by attackers, according to researchers' public alerts and the watchTowr intelligen...

AmnesiaStealer MacOS malware that steals credentials and controls real-time browser sessions
Security researchers have documented a new malware family aimed at macOS - called AmnesiaStealer - that combines a dropper in shell, an infostealer written in Rust and a remote ...